Dear Statali,

I am currently working on an event study for my thesis. The study comprises of 1834 event dates with 1554 different companies. I followed the Princeton guide for Event study (link: https://dss.princeton.edu/online_hel...ventstudy.html) in order to project normal performance and to calculate the abnormal returns per Company for different event windows. I am using the latest Stata Version.

Based on several research papers, I have to calculate the cumulated average abnormal return (CAAR) overall companies for every event window and then conduct the t-test for each CAAR at every significance level.
However, the problem I am facing right now is i) calculation of CAAR, ii) calculation of the standard deviation over all companies and iii) the execution of a t-test for the event windows +/-2, +/-5, +/-7 and +/-10 for significance levels 1%, 5% and 10%.
From my understanding, I should have only one CAAR per event window and test this value for significance.

I applied the reg formel for cumulated abnormal returns (CAR) for event window +/- 10 days using the Princeton paper method. I received the following result:

Array


However, I am not convinced that this is the right approach or right step in order to derive the different results for my thesis.
Different from the reg formula, I figured that if I summarize all abnormal returns (AR) for e.g. +/-10 days stata will provide me both the mean and standard deviation for my observations, which I can plug-in in the stata t-test calculator. the result is the following:

Array

Did any of these two methods make any sense or is there another way to generate one CAAR per event window and do a t-test for each level of significance for each event window?

Thank you in advance.
BR Luka